- Feb 13, 2023
-
-
Dylan Baker authored
-
- Feb 09, 2023
-
-
CC: mesa-stable Closes: #7987 Signed-off-by:
Tapani Pälli <tapani.palli@intel.com> Reviewed-by:
Nanley Chery <nanley.g.chery@intel.com> Part-of: <!20981>
-
The current behavior results in a fallback to building our own expat if it's not found, even in cases where it's been explicitly been disabled. This has been fixed on main, but that patch is part of a refactor that updates the required meson version, and isn't suitable for backport. Closes: #8259 Reviewed-by:
Jesse Natalie <jenatali@microsoft.com> Part-of: <!21220>
-
Point sprite coordinates in general need to be inverted, not just the texcoords converted to point sprite. Move point coord y inversion out to its own pass. Fixes GTF-GL46.gtf21.GL2FixedTests.point_sprites.point_sprites with FBO dEQP surface. cc: mesa-stable Part-of: <!21050> (cherry-picked from commit 782f1e9e)
-
- Feb 08, 2023
-
-
There was a VUID-VkImageViewCreateInfo-image-04739 in the Vulkan 1.3 spec that said: If image was created with the VK_IMAGE_CREATE_BLOCK_TEXEL_VIEW_COMPATIBLE_BIT flag and format is a non-compressed format, viewType must not be VK_IMAGE_VIEW_TYPE_3D That VUID has since been removed, and when a view of a 3D image is created, with put the depth into the array_len, so it won't be always 1. Reviewed-by:
Mark Janes <markjanes@swizzler.org> Reviewed-by:
Lionel Landwerlin <lionel.g.landwerlin@intel.com> Part-of: <!20803> (cherry picked from commit 58ababde)
-
Dylan Baker authored
-
Fixes: d8d99c3c ("aco: add GFX11 opcode numbers") Reviewed-by:
Rhys Perry <pendingchaos02@gmail.com> Part-of: <!21170> (cherry picked from commit c8adf162)
-
* drawid needs to be incremented at end of loop * rescope draw structs to initialize at top of loop fix #8191 Fixes: 6b07893b ("util/vbuf: handle multidraws") Reviewed-by:
Kenneth Graunke <kenneth@whitecape.org> Tested-by:
Mark Janes <markjanes@swizzler.org> Reviewed-by:
Marek Olšák <marek.olsak@amd.com> Part-of: <!21139> (cherry picked from commit 8f5a63eb)
-
failing to set this yields patterns like * bind fs * bind samplerviews * draw * bind fs2 * ~~unbind samplerviews~~ (eliminated) * draw the eliminated unbinding of samplerviews between draws also eliminates a descriptor update, triggering various artifacts in certain corner cases (like DOOM2016 shadows) it's possible to manage the updating during shader binding, but the detection is a bit more complex, and the cpu overhead from maintaining the current codepath with an extra pipe_context::set_sampler_views (et al) isn't high enough to warrant further investigation at this time fixes #8252 Fixes: 153af03b ("gallium: Add cap to request state validation for all dirty state") Part-of: <!21176> (cherry picked from commit b9181c32)
-
Add a parameter for cropping, seperate from the translate parameter, to the progressive compute shader. Fixes: 4c46e4a5 ("gallium/auxiliary/vl: fix scale and translate parameters") Fixes: 56eac722 ("gallium/auxiliary/vl: compute shaders for progressive yuv") Signed-off-by:
Thong Thai <thong.thai@amd.com> Tested-by:
Veerabadhran Gopalakrishnan <veerabadhran.gopalakrishnan@amd.com> Tested-by:
Suresh Guttula <suresh.guttula@amd.com> Acked-by:
Veerabadhran Gopalakrishnan <veerabadhran.gopalakrishnan@amd.com> Acked-by:
Leo Liu <leo.liu@amd.com> Part-of: <!21165> (cherry picked from commit 1fe5ffd6)
-
A few lines earlier uni_offsets is accessed with ubo scaled by PIPE_MAX_CONSTANT_BUFFERS: if (uni_offsets[ubo * PIPE_MAX_CONSTANT_BUFFERS + i] == offset) Found by inspection. Looking at the before and after NIR code for dEQP-VK.graphicsfuzz.cov-int-initialize-from-multiple-large-arrays, using the correct indexing appears to enable the pass to inline an additional uniform. My guess is that when a uniform is used more than once, the first loop wouldn't find the offset recored in the table because it was recorded at the wrong location. Fixes: d23a9380 ("lavapipe: implement extreme uniform inlining") Reviewed-By:
Mike Blumenkrantz <michael.blumenkrantz@gmail.com> Part-of: <!21144> (cherry picked from commit a7696a4d)
-
Fixes: 86367172 ("zink: add a "compact" descriptor mode") Reviewed-by:
Emma Anholt <emma@anholt.net> Part-of: <!21169> (cherry picked from commit faaf6f95)
-
this was including the generated tcs bits, which was likely to be wrong and thus break optimal key hashing, requiring more pipelines it also wasn't setting the optimal key value correctly during precompile, which meant the wrong hash value was used and the precompiled libs were never actually accessible cc: mesa-stable Reviewed-by:
Emma Anholt <emma@anholt.net> Part-of: <!21169> (cherry picked from commit 4cf54e2e)
-
this may otherwise not be set until the cache has been broken Fixes: 56fb2580 ("zink: replace mixed_zs with zs feedback loops") Reviewed-by:
Emma Anholt <emma@anholt.net> Part-of: <!21169> (cherry picked from commit bbae2372)
-
this needs to be set on context create or it may never get set Fixes: 645f2da3 ("zink: always set VK_PIPELINE_CREATE_COLOR_ATTACHMENT_FEEDBACK_LOOP_BIT_EXT sometimes") Reviewed-by:
Emma Anholt <emma@anholt.net> Part-of: <!21169> (cherry picked from commit 0cb326cc)
-
required by spec Fixes: 7ab5c5d3 ("zink: use EXT_descriptor_buffer with ZINK_DESCRIPTORS=db") Reviewed-by:
Emma Anholt <emma@anholt.net> Part-of: <!21169> (cherry picked from commit d584f320)
-
tcs values here are ignored since they only matter for generated tcs Reviewed-by:
Emma Anholt <emma@anholt.net> Part-of: <!21169> (cherry picked from commit 7c021cc5)
-
Reviewed-by:
Emma Anholt <emma@anholt.net> Part-of: <!21169> (cherry picked from commit 1f08a6dc)
-
Reviewed-by:
Emma Anholt <emma@anholt.net> Part-of: <!21169> (cherry picked from commit 77d8cd9c)
-
Dylan Baker authored
-
Even when small primitive culling is disabled, the face culling algorithm in ac_nir_cull can delete tiny triangles when their area is almost zero. Cc: mesa-stable Signed-off-by:
Timur Kristóf <timur.kristof@gmail.com> Reviewed-by:
Samuel Pitoiset <samuel.pitoiset@gmail.com> Reviewed-by:
Marek Olšák <marek.olsak@amd.com> Part-of: <!20987> (cherry picked from commit 3508597a) Conflicts: src/amd/vulkan/radv_cmd_buffer.c
-
GL Spec says that imageLoad from incomplete images must return 0. This is not really spec compliant as for proper behavior nullDescriptor and robustImageAccess2 is needed. A workaround for lack of either of these requires a shader variant. Clearing the null surface and hoping the app doesn't write to the image is closer to spec, while avoiding a shader recompile. KHR-GL46.shader_image_load_store.incomplete_textures tests this. cc: mesa-stable Part-of: <mesa/mesa!21135> (cherry picked from commit 22e91af1) Conflicts: src/gallium/drivers/zink/zink_context.c
-
u_vector_add() don't keep the returned pointers valid. After the initial size allocated in u_vector_init() is reached it will allocate a bigger buffer and copy data from older buffer to the new one and free the old buffer, making all the previous pointers returned by u_vector_add() invalid and crashing the application when trying to access it. This is reproduced when running dEQP-VK.synchronization.signal_order.timeline_semaphore.* in DG2 SKUs that has 4 CCS engines, INTEL_COMPUTE_CLASS=1 is set and of course perfetto build is enabled. To fix this issue here I'm moving the storage/allocation of struct intel_ds_queue to struct anv_queue/iris_batch and using struct list_head to maintain a chain of intel_ds_queue of the intel_ds_device. This allows us to append or remove queues dynamically in future if necessary. Fixes: e760c5b3 ("anv: add perfetto source") Reviewed-by:
Lionel Landwerlin <lionel.g.landwerlin@intel.com> Signed-off-by:
José Roberto de Souza <jose.souza@intel.com> Part-of: <mesa/mesa!20977> (cherry picked from commit 8092bc21) Conflicts: src/intel/ds/intel_driver_ds.cc src/intel/vulkan/anv_utrace.c src/intel/vulkan_hasvk/anv_utrace.c
-
- Feb 07, 2023
-
-
Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com> Reviewed-by:
Georg Lehmann <dadschoorse@gmail.com> Closes: #8224 Cc: mesa-stable Part-of: <!21138> (cherry picked from commit fad1f716)
-
this is unsupported and undefined by the spec cc: mesa-stable Reviewed-by:
Roland Scheidegger <sroland@vmware.com> Part-of: <mesa/mesa!21099> (cherry picked from commit 7d8e2db9)
-
Currently, we postpone binning syncs until we record draw calls and can validate if any of them require accessing protected resources in the binning stage, however, if the draw calls are recorded in a secondary command buffer and the barriers have been recorded in the primary command buffer, we won't apply the binning sync in the secondary when we record the draw calls and so we must apply it when we execute the secondary in the primary. Fixes flakyness in: dEQP-VK.api.command_buffers.record_many_draws_secondary_2 cc: mesa-stable Reviewed-by:
Alejandro Piñeiro <apinheiro@igalia.com> Part-of: <mesa/mesa!21162> (cherry picked from commit fec15a22)
-
The CLE parser in the sim will read this many bytes for each instruction in a CL, so we should ensure we have at least that many bytes available in the BO when reading the last instruction, otherwise we can trigger a GMP violation. It is not clear whether this behavior applies to real hardware too. cc: mesa-stable Reviewed-by:
Alejandro Piñeiro <apinheiro@igalia.com> Part-of: <!21162> (cherry picked from commit c2601f06)
-
Fixes: 46b099e3 ("meson: Ignore unused variables in release builds") 46b099e3 has some issues: - it doesn't enable unused variables warning on release builds with assertions enabled; - it doesn't disable unused variables warning on debug builds with assertions disabled; - it doesn't disable unused variables warning when building with MSVC and assertions are disabled regardless of buildtype, see #8147. 3/4 regressions reported there have this limitation alone as root cause. Reviewed-by:
Emma Anholt <emma@anholt.net> Part-of: <!21154> (cherry picked from commit 43470724)
-
This was inadvertently removed in 98a6cfd3 and causes continuous memory leaks as events are being received after the context has been unbound. Fixes: 3170b633 ("loader: Add infrastructure for tracking active CRTC resources") Closes: #8238 Signed-off-by:
Yuxuan Shui <yshuiv7@gmail.com> Reviewed-by:
Kenneth Graunke <kenneth@whitecape.org> Part-of: <!21128> (cherry picked from commit 27a89a09)
-
Dylan Baker authored
-
- Feb 06, 2023
-
-
This is the Hasvk version of Anv's: ad6a036a ("anv: don't leave undefined values in exec->syncobj_values") Reviewed-by:
Lionel Landwerlin <lionel.g.landwerlin@intel.com> Signed-off-by:
Paulo Zanoni <paulo.r.zanoni@intel.com> Part-of: <!20800> (cherry picked from commit 80196aaa)
-
This is the Hasvk version of Anv's: 3d37950f ("anv: check the return value of anv_execbuf_add_bo_bitset()") Reviewed-by:
Lionel Landwerlin <lionel.g.landwerlin@intel.com> Signed-off-by:
Paulo Zanoni <paulo.r.zanoni@intel.com> Part-of: <!20800> (cherry picked from commit 5039acfd)
-
according to spec, the maximum number of acquired images can be calculated with swapchain_size - VkSurfaceCapabilitiesKHR::minImageCount + 1 the previous calculation was both wrong and occurring in the wrong place, so this corrects both issues cc: mesa-stable Part-of: <!21095> (cherry picked from commit 89cf0a3b)
-
simple way to reproduce this is to run these 4 together: KHR-GL46.gpu_shader5.images_array_indexing KHR-GL46.shader_image_load_store.advanced-allMips KHR-GL46.shader_image_load_store.advanced-sso-simple KHR-GL46.shader_image_load_store.incomplete_textures cc: mesa-stable Part-of: <!21134> (cherry picked from commit 104040b5)
-
Found by inspection. Something probably hangs because of this, but I don't know what. Signed-off-by:
Rhys Perry <pendingchaos02@gmail.com> Reviewed-by:
Tatsuyuki Ishi <ishitatsuyuki@gmail.com> Reviewed-by:
Samuel Pitoiset <samuel.pitoiset@gmail.com> Fixes: c199a516 ("radv: bind the VS input state for prologs created with GPL") Part-of: <!20913> (cherry picked from commit 10a5035c)
-
Fixes: afd645f0 ("ac/llvm: remove LLVMBuildGEP usages") Reviewed-by:
Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com> Reviewed-by:
Samuel Pitoiset <samuel.pitoiset@gmail.com> Part-of: <!20521> (cherry picked from commit 10ac51a5)
-